This lecture delves into the powerful concept of duality in statistical physics, particularly within the context of percolation theory and resistor networks. It begins by revisiting the "animal problem" and distinguishing between clusters with and without loops, highlighting their differing growth behaviors. The core of the discussion centers on graph duality for planar lattices, demonstrating its application to derive the exact critical probability (Pc = 1/2) for bond percolation on a square lattice and site percolation on a triangular lattice due to their self-dual properties. The lecture then extends the concept to higher dimensions, noting complexities in 3D (plaquette percolation with two critical points) and self-duality in 4D. Furthermore, it explores the voltage-current duality in planar resistor networks, establishing a reciprocal relationship between conductances of dual links. Finally, the principles of duality are generalized to directed percolation, where bonds can have preferred flow directions (insulators, resistors, diodes), examining the Diode-Insulator (DI) and Diode-Resistor (DR) problem duality, and outlining the multi-phase behavior and symmetries of a general four-state directed percolation system.
Summary of Duality in Percolation and Related Problems:
0:00:05 - Animal Problem & Lattice Gas Equivalence: The lecture starts by recalling the "animal problem" (enumerating clusters), noting that directed animals are equivalent to a 1D nearest-neighbor exclusion lattice gas, and undirected animals (without loops) are equivalent to a 2D hardcore nearest-neighbor exclusion lattice gas.
0:01:00 - Animals vs. Branched Polymers: A distinction is made between "animals" (clusters with loops, AS) and "branched polymers" (clusters without loops, BS). While they share the same critical exponents (Theta_b = Theta_a), their growth constants differ (Lambda_a > Lambda_b).
0:03:51 - Introduction to Graph Duality (Planar Graphs): The fundamental concept of geometrical duality for planar graphs is introduced. In the dual graph, faces of the original graph become vertices, and edges connect dual vertices if the original faces shared a common edge. The dual of a dual graph returns the original graph.
0:10:15 - Duality in Higher Dimensions (3D & 4D): The concept of duality extends beyond 2D. In 3D, original cells correspond to dual vertices, original faces to dual edges, and original edges to dual faces. Imagining 4D duality is also possible.
0:13:23 - Bond Percolation on Square Lattice (Pc = 1/2):
The square lattice is proven to be self-dual for bond percolation.
A key insight is that if an original (red) bond is not present, its corresponding dual (blue) bond is present, and vice-versa.
Key Takeaway: In 2D, a path of original bonds connecting left-to-right cannot coexist with a path of dual bonds connecting top-to-bottom. This geometric property, combined with self-duality, leads to the exact critical probability of Pc = 1/2 for bond percolation on a square lattice (Kesten's result).
0:23:41 - Duality for Other Lattices: The triangular lattice is shown to be dual to the hexagonal lattice. This implies Pc_bond(hexagonal) = 1 - Pc_bond(triangular).
0:25:38 - Site Percolation on Triangular Lattice (Pc = 1/2): Site percolation on the triangular lattice is also argued to have Pc = 1/2, again relying on symmetry and duality principles.
0:29:09 - Duality in 3D Bond Percolation (Plaquette Percolation):
Bond percolation on a 3D cubic lattice is dual to "plaquette percolation" on a cubic lattice (where faces of cubes are occupied).
Key Takeaway (3D): Unlike 2D, in 3D, percolating original clusters (e.g., bond percolation) and percolating dual clusters (e.g., plaquette percolation) can coexist and interpenetrate. This leads to the existence of two distinct critical points for plaquette percolation (e.g., P1 ≈ 0.35 and P2 ≈ 0.65).
0:32:26 - Duality in 4D Plaquette Percolation: The 4D plaquette percolation problem is self-dual, leading to a Pc = 1/2.
A well-known "voltage-current duality" exists, linking the solution of Kirchhoff's laws using nodal voltages on the original network to using loop currents on its dual network.
Key Takeaway: For this duality to hold, the conductance (σ) of a link in the original network must correspond to a reciprocal conductance (σ' = 1/σ) in the dual network.
Introduces directed percolation, where flow is restricted to pre-specified directions (e.g., downwards or rightwards), profoundly affecting cluster behavior.
0:53:07 - Active-Absorbing State Transitions: DP is a prototype for active-absorbing state phase transitions, where a system can transition between an "active" state (e.g., spreading infection) and an "absorbing" state (e.g., infection dies out).
0:54:18 - Yensen-Grasberger Conjecture: This important conjecture posits that most systems with a unique absorbing state and no special symmetries belong to the directed percolation universality class.
0:58:39 - Duality for Directed Percolation Systems:
A more complex system is considered where bonds can be Insulators (I), Resistors (R - bidirectional flow), Diodes (D+ - positive direction), or Diodes (D- - negative direction).
1:02:22 - Diode-Insulator (DI) ↔ Diode-Resistor (DR) Duality: A specific duality transformation maps a Diode-Insulator (DI) problem to a Diode-Resistor (DR) problem. This implies that Insulator states map to Resistor states, and vice versa, while diode directions remain the same.
1:03:50 - Diode-Resistor Percolation (DRP) Behavior: In DRP, clusters formed from a single source tend to have a "staircase" shape, forming an approximate wedge with a specific angle. The critical point for DRP is related to directed percolation: Pc_DRP = 1 - Pc_DP.
1:12:59 - General (I, R, D+, D-) System Phase Diagram: The general four-state system exhibits multiple phases (no flow, positive flow, negative flow, or bidirectional flow). Its phase diagram can be visualized as a tetrahedron, possessing symmetries like Insulator ↔ Resistor exchange and D+ ↔ D- exchange.
1:23:20 - Generalized Resistor Duality for Anisotropic Conductance: The resistor duality generalizes to bonds with distinct forward (σ_F) and backward (σ_B) conductances. The dual bond conductances become their reciprocals (σ'_F = 1/σ_F, σ'_B = 1/σ_B).
1:18:25 - Exercise: A simple resistor network problem is posed, encouraging mental calculation of current flow without explicitly writing equations.